To be added

This model is an update of the earlier iteration, meant to preserve the reasoning capacities of Tess R1 Ballad and Behemoth X 2.2, which I wasn't aware of when building the earlier one.

In this version, I applied the following corrections:

(1) full-delta task arithmetic for Tess R1 on the upper half

(2) Behemoth X 2.2 + newly included Behemoth X R1 + Behemoth T1 V2 thinking adapter on the lower half.

Thank you for the 67 downloads in the earlier model (the other 3 were from me), it meant a lot. If you have any feedback or comments, please don't hesitate to chat in the Community tab. All community feedback is warmly welcomed.

Again, still not yet tested due to the lack of computational resources, I just wanted to try all variations I can come up with.


Upcoming Testing Plan

  • Using the following chat template:
[SYSTEM_PROMPT] <|system|>{{system_message}}[/SYSTEM_PROMPT][INST] <|user|>{{user_message}}[/INST] <|model|>{{assistant_message}}

Template is taken from TheDrummer/Behemoth-123B-v2.1, all credits go to the respective author(s)

Temp: 1.0, min_p: 0.03, top_p: 0.95, max_token: 204860

{{system_message}} structure:

SYSTEM

<insert your system prompt here, each section separated by a *** >

---

License

All components derive from Mistral Large 123B and are released under the Mistral AI Research License. This merge inherits that license: research and non-commercial use only. Please refer to the individual component model cards for any additional terms.


Merge recipe

Please see the recipe in the attached .yaml files.

Downloads last month
514
Safetensors
Model size
170B params
Tensor type
BF16
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for alansmithee-cpu/experimental_merge_241026_v1